← back to paper
arxiv: 2604.22891 · 2 revisions
Quantifying and Mitigating Self-Preference Bias of LLM Judges